Governing the Inferential Chain: Artificial Intelligence, Evidence, and the Logic of Sociological Inquiry

The XXI ISA World Congress highlights how AI is reshaping “the nature of knowledge and intellectual work”. This session places that transformation at the center of RC33’s agenda by asking a fundamental methodological question: what happens to sociological inquiry when key stages of knowledge production are increasingly delegated to computational systems?

The challenge posed by AI is not merely technical but epistemological and political. Machine learning models classify survey responses, large language models assist qualitative coding, generative systems synthesize documents, and algorithmic infrastructures increasingly mediate the production of evidence itself. Yet the inferential chain connecting computational outputs to sociological claims often remains opaque. While current debates oscillate between enthusiasm for new analytical capacities and concerns about ethical risks, less attention has been devoted to understanding how algorithmic outputs become evidence and how evidence becomes sociological knowledge.

This question is particularly relevant in a context where knowledge production is shaped by inequalities in access to data, computational resources, proprietary models, and research infrastructures. AI is not only transforming methods; it is also redistributing epistemic authority and redefining who can produce, validate, and govern knowledge.

Three dimensions structure the session: technical validation (on quality, robustness, and reproducibility of AI-generated outputs); interpretive construction (on translation of algorithmic patterns into sociological meaning); epistemic governance (on accountability, transparency, and power relations in automated research workflows).

Empirical studies, theoretical frameworks, methodological innovations, and critical reflections in the light of comparative, global, and Global South perspectives on AI, evidence, and research governance are especially welcome.